This invention relates to solid state light sources based on LEDs mounted on or within thermally conductive luminescent elements, which provide both convective and radiative cooling. Low cost, self-cooling solid state light sources can be realized by also integrating the electrical interconnect of the LEDs and other semiconductor devices. The thermally conductive luminescent element can be used to completely or partially eliminate the need for any additional heatsinking means by efficiently transferring and spreading out the heat generated in LED and luminescent element itself over an area sufficiently large enough such that convective and radiative means can be used to cool the device.
Solid luminescent ceramic plates are disclosed in Born, U.S. Pat. No. 4,849,639, as scintillators for the conversion of shorter wavelength photons to longer wavelength photons. Born does not disclose the use of the ceramic plates to convert narrowband emission from solid state emitters into broadband visible light sources.
CeYag in ceramic, single crystal, and powder forms have been used since the early 1970s to convert blue wavelengths to yellow wavelengths for efficient coupling of flashlamp outputs into laser rods at Bell Labs and other institutions. Again, the use of these materials to convert blue/UV solid state emitters into broadband white light sources is not disclosed.
The formation of transparent and translucent luminescent YAG ceramics are disclosed in Cusano. U.S. Pat. No. 4,421,671, and Greskovich, U.S. Pat. No. 4,466,930. The use of these materials as scintillators, CRT faceplates, and X-ray screens is disclosed but not their use in solid state lighting.
The use of solid wavelength conversion elements to form self cooling light sources in which the solid wavelength conversion element is a ceramic, single crystal, composite and layered solid material is disclosed in Zimmerman, U.S. Pat. No. 7,285,791. Zimmerman discloses the use of solid wavelength conversion elements to form solid state light sources in U.S. Pat. No. 7,804,099. The use of high emissivity surfaces is not disclosed.
Mueller-Mach et al. in U.S. Pat. No. 6,696,703 disclose the deposition of a thin film phosphor directly on the LED die. However, as-deposited thin film phosphors have relatively poor wavelength conversion efficiency. A high-temperature annealing step is required in order to properly activate the phosphor. This annealing step can damage the semiconductor layers of the LED. In addition, the absorption cross-sections of most thin film phosphors are low, especially for blue and near ultraviolet (UV) excitations typically used within solid-state lighting. It is neither economical nor practical in most cases to create a sufficiently thick layer of luminescent material directly on the LED. Another drawback to depositing a phosphor directly on the LED die is that a large portion of the light generated within a deposited phosphor layer can be trapped due to total internal reflectance. The need therefore exists for a method to utilize high performance phosphors within an LED package such that the best phosphor can be used efficiently (e.g. with sufficient quantity, minimal backscatter, and maximum light extraction). The need also exists for a method to fabricate high efficiency phosphors without damaging the LED semiconductor layers. In addition, high emissivity is not disclosed.
Another important aspect of phosphors relates to characterization and overall device efficiency. Phosphors are typically characterized in terms of quantum efficiency and Stokes shift losses. As an example, a powder phosphor layer is deposited on a glass surface and excited. The luminescence is measured as a function of excitation energy and the result is usually compared to a standard phosphor of known quantum efficiency. The losses associated with Stokes shift can be subtracted and the result would be the intrinsic quantum efficiency. Several problems exist with this method of characterization such as backscattered light, coating thickness variability and light trapping. In the case of phosphor powders, the majority of the generated light can escape from the phosphor particles due to their substantially spherical nature and to scattering centers located on or in the material itself. The main problem measuring the efficiency of phosphor powders is backscattering of the light from thick samples. For deposited phosphor films or grown phosphor boules, however, the problem of measuring the phosphor efficiency is affected by light extraction. The majority of the light generated in the phosphor can be trapped within the material itself due to total internal reflection. Several approaches have been used to solve the total internal reflection problem including various roughening techniques and shaping approaches. In these cases, the overall efficiency becomes as much a function of the extraction means as the conversion efficiency. Deposited phosphor films have the added complication of a secondary substrate material with its associated indices and losses.
Mayer et al. in U.S. Pat. No. 6,565,770 describe thin interference pigment flakes that can be made on a flexible substrate and then mechanically removed by flexing the substrate. The dichroic reflectors discussed are used in security enhancement on money and other decorative optical effects. The use of luminescent materials is discussed but is related to the formation of a particular optical effect such as UV illumination for security markings. No explanation for improving the output efficiency of LEDs or other light emitting devices is discussed and no device based on integrating the phosphor layer with the excitation source to form an efficient solid-state lighting element is disclosed.
The use of flake-like phosphors is also discussed by Aoki et al. in U.S. Pat. No. 6,667,574 for use in plasma displays, but the patent again lacks any reference to solid-state lighting applications or methods to enhance their output. In addition, the above two applications are very much cost driven because of the large areas typically required in making a plasma display or the marking of money or decorative items. In order to maximize the performance of these wavelength-converting materials high temperature processing is preferred.
Mueller-Mach et al. in U.S. Pat. No. 6,630,691 disclose a thin single-crystal phosphor substrate onto which an LED structure is fabricated by epitaxial growth techniques. However, single-crystal phosphor substrates are expensive and finding a single crystal phosphor substrate that has the proper lattice match to allow the growth of the LED structure can be difficult.
Ng et al. in US Patent Application No. 20050006659 disclose a planar sheet of a single-crystal phosphor that is placed over the output surface of an LED as a portion of a preformed transparent cap. However, single-crystal phosphor sheets must be grown by epitaxial processes or sliced from bulk single crystals of phosphor material. Single crystal phosphor sheets are therefore too expensive for most practical applications. Planar sheets of polycrystalline phosphors are not disclosed in US Patent Application No. 20050006659. Bonding the planar sheet of a single-crystal phosphor directly to the surface of the LED to improve heat dissipation in the phosphor sheet is also not disclosed.
LEDs unlike conventional light sources such as incandescent bulbs cannot effectively cool themselves. As such additional heatsinking or cooling means are required to prevent overheating. This increases the cost of not only the light sources due to shipping costs and materials costs but also the fixtures that use those light sources. In general, the need exists for articles and means which allow LEDs to be used without the need for additional heatsinking means.
It is desirable to minimize the temperature difference between the junction or active region of the semiconductor device and the ambient atmosphere to effectively cool small semiconductor devices. It is also desirable to minimize the surface area needed to dissipate the heat generated by the semiconductor devices to the ambient atmosphere. While high thermal conductivity materials can be used to spread the heat out over a very large area, these high thermal conductivity materials come with the addition of significant weight and cost. In conventional LED devices several layers of interconnect exist between the LED die and the final light source. This approach is used because the lighting fixture manufacturers have historically not been required or had the capability to wirebond, flipchip attach or even solder components into their fixtures. Also the need to regularly replace light sources such as incandescent bulbs has led to a wide range of quick change interconnects like sockets and pin based connector. Lightweight self cooling solid state light sources would offer significant benefits to fixture manufacturers. Incandescent bulbs for instance are very lightweight generating over 1000 lumens while weighing only 50 grams and as such can be easily held in place using even simple pins and sockets. For the typical LED sources, this is not the case. The added weight of the heatsink and the need for a low resistance thermal connection between the LED package and the heatsink necessitates the use of complex multiple level interconnects. The need exists for LED light sources which are lightweight and easily incorporated into a wide range of lighting fixtures without the need for additional heatsinking or cooling means.
As is well known to those skilled in the art, phosphor conversion is typically used to broaden the narrow band emission of LEDs to more closely match the sun or incandescent spectrum. This is usually done via phosphor powders mixed into an organic matrix. Using this conventional approach, the heat generated in the phosphor powders is thermally isolated from the ambient by the organic matrix.
For example, blue InGaN LEDs are routinely coated with a thin organic layer containing phosphor powders. The organic material typically consists of a silicone or epoxy. As the LED efficiency and flux density has increased, more of the thermal losses are localized in the phosphor powders. Unfortunately, thermal conductivity of the luminescent layer is mainly determined by the thermal conductivity of the organic matrix material, which is typically around 0.1 W/m/K. Typically a 50 micron coating thickness for the luminescent organic layer prevents high scatter losses created by the index of refraction difference between the phosphor powder and organic matrix material. Conversely sufficient phosphor powder must be used to convert the shorter wavelength excitation to longer wavelength emission this is typically controlled by the loading level which is the ratio of the percentage of phosphor powder to the percentage of organic matrix material. A lower loading level of the phosphor powder reduces the thermal conductivity and a higher loading level limits the thickness of coating due to scatter losses.
Generically it is difficult to remove a significant amount of heat out of the phosphor powders, let alone the LEDs themselves, while they are in a low thermal conductivity matrix. Their luminescent efficiency typically decreases as the phosphor powders get hotter. This luminescent inefficiency has spurred the development of remote phosphor approaches which reduce the thermal load on the phosphor powders by moving the phosphor powder farther away from the LED and thereby reducing the flux density per unit area on the powder. This remote phosphor approach however increases the source size, amount of phosphor powder required, forms a thermal barrier around the LEDs, and creates a large volume light source.
In addition, organic systems are susceptible to blue and UV radiation due to the photostability of the C—H bonds which define an organic system. Photostabilization of especially transparent optical systems under intense solarization has limited the long-term use of transparent organics (plastics) in greenhouse and other outdoors applications. The solar constant is approximate 1000 optical watts per square meter, with less than 10% of that having a wavelength short enough to photochemically attack the C—H bond. A typical blue LED in solid state lighting applications will output up to 1 optical watt per square millimeter of which virtually all the wavelengths emitted are capable of adversely affecting the C—H bonds of organic materials. These irradiation levels represent four order of magnitude higher flux densities than greenhouse films experience. Accordingly, inorganic solutions are more desirable than organic solutions for thermal conductivity and photostability standpoints.
All wavelength conversion materials and semiconductor devices exhibit temperature dependent efficiency curves. Thermal roll-off occurs for Ce:YAG around 150 degrees C. Alternately, AlinGaP red diodes and InGaN blue die both exhibit some roll-off as the junction temperature exceeds 100 degrees C. It therefore becomes critical that the heat generated within a solid state lighting system is transferred to the surrounding ambient using the lowest possible thermal resistance path. In the case of natural convection cooling the amount of heat that can be transferred to the surrounding ambient (air) is determined by the natural convection heat transfer coefficient, the area of the surface, and the temperature of the surface relative to the surrounding air or ambient.
Radiative cooling can also contribute to cooling solid state lighting devices if the temperature difference between the junction temperature and the radiating surface is minimized over a sufficiently large high emissivity surface area of the lighting device. Proposed solutions such as forced convection cooling, heatpipes, and even liquid cooling, either move the heat around or substantially increase the volume and weight of the light source. These solutions result in very low lumens/gram light sources.
Historically, light sources have cooled themselves as stated earlier. In the case of incandescent and fluorescent tubes, the glass envelope surrounding the sources, and the filament or arc itself transfers the excess heat generated via convection and radiation. An incandescent bulb glass envelope can exceed 150 degrees C. and a halogen's quartz envelope may exceed several hundred degrees celsius. Radiative power scales as the fourth power of the temperature. A naturally convectively cooled surface with a surface temperature of 50 degrees C. in a 25 degrees C. ambient will transfer only about 5% of its energy to the surrounding ambient radiatively. A naturally convectively cooled surface with a surface temperature of 100 degrees C. can transfer 20% of its energy to the surrounding ambient radiatively. The typical LED junction temperature for high powered devices can be over 120 degrees C. and still maintain excellent life and efficiency. For surfaces with temperatures less than 120 degrees C. the majority of the radiated energy is in the infrared with a wavelength greater than 8 microns.
The emissivity of the materials in the infrared varies greatly. Glass has an emissivity of approximately 0.95 while aluminum oxide may be as low as 0.23. Organics such as polyimides can have fairly high emissivity in thick layers. This however will negatively affect the transfer of heat due to the low thermal conductivity of organics.
In order to maximize heat transfer to the ambient atmosphere, the need exists for luminescent thermally conductive materials which can effectively spread the heat generated by localized semiconductor and passive devices (e.g. LEDs, drivers, controller, resistors, coils, inductors, caps etc.) to a larger surface area than the semiconductor die via thermal conduction and then efficiently transfer the heat generated to the ambient atmosphere via convection and radiation. At the same time, these luminescent thermally conductive materials must efficiently convert at least a portion of the LED emission to another portion of the visible spectrum to create a self cooling solid state light source with high L/W efficiency and good color rendering.
Heat generated within the LEDs and phosphor material in typical solid state light sources is transferred via conduction means to a much larger heatsink usually made out of aluminum or copper. The temperature difference between the LED junction and heatsink can be 40 to 50 degrees C. The temperature difference between ambient and heatsink temperature is typically very small given the previously stated constraints on the junction temperatures of LEDs. This small temperature difference not only eliminates most of the radiative cooling but also requires that the heatsink be fairly large and heavy to provide enough surface area to effectively cool the LEDs. This added weight of the heatsink increases costs for shipping, installation and in some cases poses a safety risk for overhead applications.
Ideally, like incandescent, halogen, sodium, and fluorescent light sources, the emitting surface of the solid-state light source would also be used to cool the source. Such a cooling source would have an emitting surface that was very close to the temperature of the LED junctions to maximize both convective and radiative cooling. The emitting surface should be constructed of a material that exhibited sufficient thermal conductivity to allow for the heat from small but localized LED die to be spread out over a sufficiently large enough area to effectively cool the LEDs.
This invention discloses thermally conductive luminescent elements within solid state light sources which overcome these issues.